Skip to content

Translation in progress Some entries still include text in the original language.

Oct 12 · Monday

All papers

Found 22 papers

Another 240papers do not have category tags yet and are not included in filtered results.

  1. AttacksarXiv 2610.09027

    P-VMI:对抗图像经 KV cache 持续影响多轮对话

    提出持久视觉记忆注入,使对抗图像经缓存在触发时操控输出

    Published
    Evaluated models
    Qwen3-8B, Qwen3-VL-8B-Instruct, Gemma-4-12B

    Abstract论文提出 P-VMI,证实对抗输入被掩蔽后仍可经 KV 缓存持久控制模型,强调了保留状态的安全风险。

    Paper analysisRead analysis
  2. AttacksarXiv 2605.21541

    研究者提出 FRA-Attack 频域正则化方法,提升对闭源 MLLM 的迁移攻击效果

    提出 FRA-Attack,用频域对齐与梯度正则做可迁移定向视觉攻击

    Published
    Attacker
    Black-box
    Evaluated models
    GPT-5.2, GPT-5.4, Claude-Opus-4.6 and others15
    AbstractFRA-Attack 以频域对齐和梯度低通做可迁移定向攻击,闭源面板上作者称整体更强,Gemma 除外。

    FRA-Attack 是面向闭源 MLLM 的迁移式定向图像攻击。作者要解决的是:空间域局部对齐重复全局低频,代理梯度的高频成分又把更新拉向代理自身。

    Paper analysisRead analysis
  3. AttacksarXiv 2605.07399

    研究者提出 GPO-V,用全局概率优化越狱扩散视觉语言模型

    提出 GPO-V,用全局概率优化的视觉扰动越狱扩散视觉语言模型

    Published
    Evaluated models
    LLaDA-V, LaViDA
    Abstract作者用视觉扰动改 dVLM 的全局生成概率,以绕过两类拒绝。

    GPO-V 把越狱从“开头必须顺从”改成“整段去噪概率朝肯定方向走”。。

    Paper analysisRead analysis
  4. AttacksarXiv 2604.14604

    The AudioHijack framework achieves covert audio prompt injection against 13 speech large models

    提出 AudioHijack,用不可感知对抗音频对语音模型做上下文无关提示注入

    Published
    Evaluated models
    SpeechGPT, GLM-4-Voice, VITA-Audio and others15

    Abstract提出了 AudioHijack 框架,揭示了 LALM 对听觉提示注入的脆弱性与传统防御的局限。

    Paper analysisRead analysis
  5. AttacksarXiv 2604.03117

    Unveiling Physical-World Semantic Vulnerabilities: Universal Adversarial Patches for Infrared Vision-Language Models

    Proposes the universal curved mesh patch UCGP, causing cross-task semantic bias in infrared vision-language models.

    Published
    Attacker
    Gray-box
    Evaluated models
    OpenAI CLIP ViT-L/14, OpenCLIP ViT-B/16, Meta-CLIP ViT-L/14 and others10
    AbstractUCGP uses one shared curved-grid patch to simultaneously lower the classification and open-ended outputs of IR-VLM.

    UCGP is a universal physical adversarial patch for infrared vision-language models. The question is: can a patch that can be attached and reused across samples cause stable semantic deviations in the infrared visual evidence shared by classification, description, and VQA. The patch is constrained to a low-frequency curved mesh.

    Paper analysisRead analysis
  6. AttacksarXiv 2602.09431

    Grounding-driven attack: Improving the transferability of encoder-based adversarial attacks to large vision-language models

    Propose GDA, which aligns image adversarial perturbations with text-grounded evidence and transfers attacks to vision-language models

    Published
    Evaluated models
    LLaVA, Qwen2.5-VL, InternVL3 and others10
    AbstractGDA aligns perturbations with text-grounded evidence, and the authors say this improves zero-query black-box transferability on heterogeneous LVLMs.

    GDA is a zero-query black-box encoder transfer attack for heterogeneous LVLMs.

    Paper analysisRead analysis
  7. AttacksarXiv 2602.01574

    SGHA-Attack: Semantically Guided Hierarchical Alignment for Transferable Targeted Attacks on Vision-Language Models

    提出 SGHA-Attack,对视觉语言模型做可迁移定向图像扰动

    Published
    Evaluated models
    UniDiffuser, BLIP-2, InstructBLIP and others9
    Abstract用多锚点与分层对齐做定向迁移。

    SGHA-Attack 是对视觉语言模型的可迁移定向攻击:只在 CLIP 代理上优化扰动,让黑盒模型按指定文本描述图像。要处理的困难是,单一参考加上只对齐代理末层,会使优化绑在该嵌入空间上,中间层语义用不上,异构 VLM 上的指定描述更容易失败。引言把绕过安全护栏列为动机,测的则是描述是否被导向目标文本。

    Paper analysisRead analysis
  8. AttacksarXiv 2601.21531

    Adversarial robustness of large vision-language models under visual token compression

    提出 CAGE 攻击,在视觉 token 压缩下用图像扰动使预测偏离

    Published
    Attacker
    Gray-box
    Evaluated models
    LLaVA, Qwen2.5-VL, InternVL3-8B and others5
    AbstractCAGE 对齐压缩瓶颈后,压缩 LVLM 的稳健准确率低于全 token 编码器攻击。

    CAGE 研究的是:LVLM 在推理时剪枝或合并视觉 token 之后,对抗稳健性该怎么估。

    Paper analysisRead analysis
  9. AttacksarXiv 2601.06464

    Adversarial Robustness of 3D Large Vision-Language Models

    提出 Vision 与 Caption 攻击,扰动点云操纵 3D VLM 输出

    Published
    Evaluated models
    PointLLM-7B, GPT4Point-2.7B
    Abstract作者用两类白盒攻击评估 3D VLM,称无目标更易成功、定向更难。

    对 point-based 3D VLM 的白盒对抗鲁棒性评测。 作者关心驾驶、机器人等场景里,小几何扰动能否改写类别或描述,以及定向诱导特定输出是否像 2D 那样容易。方法上,Vision Attack 用余弦相似度推远原视觉 token,或拉向另一点云的特征。

    Paper analysisRead analysis